Vugar Ismayilov and Namig Guliyev, employees of the Institute of Mathematics and Mechanics, published the article “On the approximation by single hidden layer feedforward neural networks with fixed weights” in “Neural Networks” magazine, the official journal of the International, European and Japanese Neural Networks Societies in 2018. the number of references has reached 100 in the Google Scholar database. 95 of the references were made by various foreign experts. The article can be found on the following page:

https://doi.org/10.1016/j.neunet.2017.12.007

It is shown in the article that a single-layer hidden-layer neural network with fixed weights, with only 2 neurons in its hidden layer, can approximate any univariate continuous function on any compact set with arbitrary accuracy.

Previous Next

© All rights are reserved. Citing to www.imm.az is necessary upon using news.